Insights metrics and data

  

Understand the data you’re getting with Insights powered by Brandwatch, how each mention gets annotated, and how metrics are calculated.

Metrics

Every quick search and saved search generates a standard set of metrics, so you can see at a glance how your search results stack up. The metrics update according to the selected time period, as well as any applied filters.

The time zone for all search data is UTC.

The following metrics display for each search:

Total Mentions The total number of mentions found in the search.
Mention Volume Over Time A line chart showing the volume of mentions per day during the selected time period.
Hover your mouse along the graph to see the exact number of mentions per day.
Word Cloud A word cloud representing the most frequently used terms and phrases within the mentions. Identify trends and themes in the search, and the language your audience uses. The bigger the word, the more frequently it occurs relative to the other terms.
Click on a word/phrase to see all the mentions that contain it.
Sentiment Volume A pie chart showing the breakdown of the results by positive, neutral, and negative sentiment.
Hover your mouse over each segment to view the exact number of mentions in each.
Sentiment Volume Over Time A line chart showing the volume of mentions for each sentiment per day during the selected time period.
Hover your mouse along the graph to see the exact number of mentions per day.
Emotion Volume A bar chart showing the volume of mentions broken down by emotion.
Hover your mouse over each bar to see exact numbers.
Emotions Volume Over Time A line chart showing the volume of mentions for each emotion per day during the selected time period.
Hover your mouse along the graph to see the exact number of mentions per day.
Top Shared URLs A list of the top 10 shared links on Twitter, sorted by the frequency that they appear in mentions.
Click a link to view all the mentions that contain it.
Top Sites A list of the top domains in the results where the mentions occur (all mentions have a URL), sorted by frequency of mentions.
Click one to view only the mentions coming from that domain.
Top Authors A list of the top 10 authors across any platform for which we have a populated author field (usually Twitter, Reddit, and Tumblr), sorted by frequency of mentions.
Top Hashtags A list of the top 10 most used hashtags in the search’s Twitter mentions, sorted by frequency.
Click on a hashtag to view a list of those mentions.
Top Interests A list of the top 10 interests associated with the authors of the mentions returned by the query. They are sorted by mention frequency (as opposed to unique author frequency).
Top Languages A list of the top languages found in the query results, sorted by number of mentions, and listing the number of unique authors found writing in each language.
Gender Split A statistical approximation of mentions by binary gender, as a percentage, based on author name.
Volume by countries A list of the top 10 countries in the search results, sorted by mention frequency, and listing the number of unique authors found from each country. The results are also displayed on a world map, with gradients of colour for each country represented. The darker the colour, the more frequently the country occurs in the results.
Click a country to view only those mentions.

Mention sources

Hootsuite Insights, powered by Brandwatch pulls mentions from the following platforms:

  • Twitter
  • Facebook
  • Instagram
  • Reddit
  • Tumblr
  • Mumsnet
  • News
  • Blogs (includes Disqus)
  • Forums
  • Reviews
  • Discuz
  • YouTube

Search limits

Quick search mentions go back 30 days, while saved searches are backfilled 30 days, and then collect and store the last 13 months of mentions.

There is no limit to the total number of mentions stored in a search. But, sampling is applied to searches that produce a high volume of mentions. For the majority of queries, sampling is not required.
When a search's 85th percentile for daily volume is greater than 5K mentions, it is considered high volume.

When a new search is created, we measure the daily volume for the past 30 days. If at least 85% of those days have a daily volume below 5K, all mentions are shown, even if the volume spikes above 5K in the future.

If more than 15% of those days have a daily volume above 5K, a sampling rate is applied to keep 85% of the daily stored mentions at or below 5K. For example, if the 85th percentile for daily volume is 10K, a 50% sampling rate is applied, so days with a true volume of 60K will have a 30K sample, and days with a true volume of 10K will have a 5K sample. All metrics for the query are extrapolated based on the sample.

Regardless of the fixed sample rate, no more than 100K mentions are stored for a single day.

Annotations

Emotion and sentiment are applied to mentions using custom statistical classifiers built in-house by Brandwatch. The classifiers learned to predict emotion and sentiment based on a training set of millions of pieces of data, including words, phrases, emojis, and other patterns.

Insights predicts sentiment in Arabic, Chinese, Czech, Dutch, English, Farsi, French, German, Greek, Hebrew, Indonesian, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Romanian, Russian, Spanish, Swedish, Thai, and Turkish.

Emotion is only supported for English mentions.

Location data is sometimes provided directly by the platform and/or the author of the mention. When an author or mention has no explicit location, we attempt to infer a location based on the keywords in the author’s bio.

Gender is a statistically informed inference based on the binary gender distribution of author names in census data and other public records.

Supported languages

Mentions in Insights powered by Brandwatch are collected in the following languages. See above for a list of languages that support emotion and sentiment annotations.

Arabic, Azerbaijani, Basque, Bulgarian, Catalan, Chinese, Croatian, Czech, Danish, Dutch, English, Farsi, Finnish, French, German, Greek, Hebrew, Hindi, Hungarian, Indonesian, Italian, Japanese, Kazakh, Korean, Kurdish, Lithuanian, Malay, Norwegian, Polish, Portuguese, Romanian, Russian, Serbian, Slovak, Slovene, Spanish, Swedish, Thai, Turkish, Urdu

In this article
Was this article helpful?
0 out of 0 found this helpful